Should manufacturers build their own AI tools? The real cost
The question is spreading through packaging and label manufacturing plants: if AI can write code in plain English, why pay for software at all? Plant managers and ops teams are now building scripts, automations, data pipelines, and internal tools using AI coding assistants, doing something that was essentially impossible two years ago.
HiFlow Solutions, a packaging ERP vendor with 25 years in the space, makes the case that the real costs of DIY AI show up after the code ships, not before.
What Claude Code actually does
Claude Code is not a chatbot. It’s an agentic coding assistant that runs in a terminal and writes or modifies software based on plain-language instructions. You give it a task, not a question: connect this system to that system, build a dashboard, summarize customer activity. For clean data and well-defined business rules, the results can work surprisingly well.
Packaging operations, HiFlow argues, are neither clean nor well-defined.
The context problem
The workflows inside a packaging plant were built over years by people who know that a specific customer’s substrate requirement interacts with a specific press configuration in ways that only show up when something goes wrong. AI coding tools don’t carry that institutional knowledge. They only know what you explicitly tell them.
HiFlow lists the variables a DIY tool won’t know unless trained on them specifically: substrate pricing by grade, supplier, and contract period; gang run logic tied to press configuration and customer approval restrictions; waste factors that vary by die configuration; and color tolerances that interact with press and substrate combinations.
The hidden labor equation
The code might generate in minutes. Confidence in that code does not. HiFlow outlines the labor the DIY math often skips:
- Defining all business rules up front
- Teaching the AI your specific operations
- Testing every exception condition
- Validating results against real data
- Documenting the logic
- Securing and protecting sensitive customer data
- Maintaining the application over time
- Building a process for fixing AI-generated code when it breaks
- Updating it as the business changes
All of that is labor cost. HiFlow says it belongs in every ROI calculation for a DIY AI project.
⚠️ The margin risk
Packaging manufacturing runs on thin margins. A substrate cost miscalculation that rounds incorrectly across high-volume runs, a waste factor that misses a die configuration, an AP entry that misclassifies a materials cost, an approval step that a workflow script quietly bypasses because the exception was never defined. Any one of those can distort a job profitability report before anyone notices.
Tech Radar’s coverage of agentic AI risks frames the broader pattern:
“Among the risks agentic AI poses, ‘shadow AI’ has emerged as a consequence of employees using unauthorized, unsanctioned AI tools or applications. When proper IT oversight or approval gets bypassed, it sets the stage for noncompliance and reputational damage.”
Who actually maintains it
In most midsize operations, the person building the DIY AI tool is a motivated production manager skilled in Excel or an IT generalist who also handles the network and the printers. AI coding tools make it easier to create something that works on a good day. The harder problem is building something that still works when the unexpected happens, which in manufacturing is most days.
HiFlow’s core argument: a purpose-built packaging ERP has already encountered, logged, and fixed the edge cases a DIY tool will meet for the first time during a production run at 2am, with no support contract to call.
The practical takeaway for any operator considering DIY AI in a manufacturing context: the code is the easy part. The governance, maintenance, and exception handling are where the real hours go.
I’m founder of BizStack at Cagri Sarigoz LLC and a passionate advocate for entrepreneurs.
With over 14 years in tech, marketing, and AI, including my role as Head of SEO at CitizenShipper and co-founder of TaleBot at Intale AI, I’m dedicated to sharing genuine, useful product insights and tips.
At BizStack, I aim to cut through the digital noise to provide clear, actionable advice.
And more than all else, I’m a father to a (always) little girl and a husband.
Contact me at [email protected] for assistance.
